xref: /linux-6.15/include/net/ip.h (revision a22b8f4b)
1 /*
2  * INET		An implementation of the TCP/IP protocol suite for the LINUX
3  *		operating system.  INET is implemented using the  BSD Socket
4  *		interface as the means of communication with the user level.
5  *
6  *		Definitions for the IP module.
7  *
8  * Version:	@(#)ip.h	1.0.2	05/07/93
9  *
10  * Authors:	Ross Biro
11  *		Fred N. van Kempen, <[email protected]>
12  *		Alan Cox, <[email protected]>
13  *
14  * Changes:
15  *		Mike McLagan    :       Routing by source
16  *
17  *		This program is free software; you can redistribute it and/or
18  *		modify it under the terms of the GNU General Public License
19  *		as published by the Free Software Foundation; either version
20  *		2 of the License, or (at your option) any later version.
21  */
22 #ifndef _IP_H
23 #define _IP_H
24 
25 #include <linux/types.h>
26 #include <linux/ip.h>
27 #include <linux/in.h>
28 #include <linux/skbuff.h>
29 
30 #include <net/inet_sock.h>
31 #include <net/snmp.h>
32 #include <net/flow.h>
33 
34 struct sock;
35 
36 struct inet_skb_parm {
37 	struct ip_options	opt;		/* Compiled IP options		*/
38 	unsigned char		flags;
39 
40 #define IPSKB_FORWARDED		1
41 #define IPSKB_XFRM_TUNNEL_SIZE	2
42 #define IPSKB_XFRM_TRANSFORMED	4
43 #define IPSKB_FRAG_COMPLETE	8
44 #define IPSKB_REROUTED		16
45 
46 	u16			frag_max_size;
47 };
48 
49 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
50 {
51 	return ip_hdr(skb)->ihl * 4;
52 }
53 
54 struct ipcm_cookie {
55 	__be32			addr;
56 	int			oif;
57 	struct ip_options_rcu	*opt;
58 	__u8			tx_flags;
59 };
60 
61 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
62 
63 struct ip_ra_chain {
64 	struct ip_ra_chain __rcu *next;
65 	struct sock		*sk;
66 	union {
67 		void			(*destructor)(struct sock *);
68 		struct sock		*saved_sk;
69 	};
70 	struct rcu_head		rcu;
71 };
72 
73 extern struct ip_ra_chain __rcu *ip_ra_chain;
74 
75 /* IP flags. */
76 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
77 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
78 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
79 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
80 
81 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
82 
83 struct msghdr;
84 struct net_device;
85 struct packet_type;
86 struct rtable;
87 struct sockaddr;
88 
89 int igmp_mc_proc_init(void);
90 
91 /*
92  *	Functions provided by ip.c
93  */
94 
95 int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
96 			  __be32 saddr, __be32 daddr,
97 			  struct ip_options_rcu *opt);
98 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
99 	   struct net_device *orig_dev);
100 int ip_local_deliver(struct sk_buff *skb);
101 int ip_mr_input(struct sk_buff *skb);
102 int ip_output(struct sk_buff *skb);
103 int ip_mc_output(struct sk_buff *skb);
104 int ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
105 int ip_do_nat(struct sk_buff *skb);
106 void ip_send_check(struct iphdr *ip);
107 int __ip_local_out(struct sk_buff *skb);
108 int ip_local_out(struct sk_buff *skb);
109 int ip_queue_xmit(struct sk_buff *skb, struct flowi *fl);
110 void ip_init(void);
111 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
112 		   int getfrag(void *from, char *to, int offset, int len,
113 			       int odd, struct sk_buff *skb),
114 		   void *from, int len, int protolen,
115 		   struct ipcm_cookie *ipc,
116 		   struct rtable **rt,
117 		   unsigned int flags);
118 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
119 		       struct sk_buff *skb);
120 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
121 		       int offset, size_t size, int flags);
122 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
123 			      struct sk_buff_head *queue,
124 			      struct inet_cork *cork);
125 int ip_send_skb(struct net *net, struct sk_buff *skb);
126 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
127 void ip_flush_pending_frames(struct sock *sk);
128 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
129 			    int getfrag(void *from, char *to, int offset,
130 					int len, int odd, struct sk_buff *skb),
131 			    void *from, int length, int transhdrlen,
132 			    struct ipcm_cookie *ipc, struct rtable **rtp,
133 			    unsigned int flags);
134 
135 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
136 {
137 	return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
138 }
139 
140 /* datagram.c */
141 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
142 
143 void ip4_datagram_release_cb(struct sock *sk);
144 
145 struct ip_reply_arg {
146 	struct kvec iov[1];
147 	int	    flags;
148 	__wsum 	    csum;
149 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
150 				/* -1 if not needed */
151 	int	    bound_dev_if;
152 	u8  	    tos;
153 };
154 
155 #define IP_REPLY_ARG_NOSRCCHECK 1
156 
157 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
158 {
159 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
160 }
161 
162 void ip_send_unicast_reply(struct net *net, struct sk_buff *skb, __be32 daddr,
163 			   __be32 saddr, const struct ip_reply_arg *arg,
164 			   unsigned int len);
165 
166 struct ipv4_config {
167 	int	log_martians;
168 	int	no_pmtu_disc;
169 };
170 
171 extern struct ipv4_config ipv4_config;
172 #define IP_INC_STATS(net, field)	SNMP_INC_STATS64((net)->mib.ip_statistics, field)
173 #define IP_INC_STATS_BH(net, field)	SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
174 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
175 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
176 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
177 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
178 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
179 #define NET_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
180 #define NET_INC_STATS_USER(net, field) 	SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
181 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
182 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
183 
184 unsigned long snmp_fold_field(void __percpu *mib[], int offt);
185 #if BITS_PER_LONG==32
186 u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t sync_off);
187 #else
188 static inline u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t syncp_off)
189 {
190 	return snmp_fold_field(mib, offt);
191 }
192 #endif
193 int snmp_mib_init(void __percpu *ptr[2], size_t mibsize, size_t align);
194 
195 static inline void snmp_mib_free(void __percpu *ptr[SNMP_ARRAY_SZ])
196 {
197 	int i;
198 
199 	BUG_ON(ptr == NULL);
200 	for (i = 0; i < SNMP_ARRAY_SZ; i++) {
201 		free_percpu(ptr[i]);
202 		ptr[i] = NULL;
203 	}
204 }
205 
206 extern struct local_ports {
207 	seqlock_t	lock;
208 	int		range[2];
209 } sysctl_local_ports;
210 void inet_get_local_port_range(int *low, int *high);
211 
212 extern unsigned long *sysctl_local_reserved_ports;
213 static inline int inet_is_reserved_local_port(int port)
214 {
215 	return test_bit(port, sysctl_local_reserved_ports);
216 }
217 
218 extern int sysctl_ip_nonlocal_bind;
219 
220 /* From inetpeer.c */
221 extern int inet_peer_threshold;
222 extern int inet_peer_minttl;
223 extern int inet_peer_maxttl;
224 
225 /* From ip_input.c */
226 extern int sysctl_ip_early_demux;
227 
228 /* From ip_output.c */
229 extern int sysctl_ip_dynaddr;
230 
231 void ipfrag_init(void);
232 
233 void ip_static_sysctl_init(void);
234 
235 static inline bool ip_is_fragment(const struct iphdr *iph)
236 {
237 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
238 }
239 
240 #ifdef CONFIG_INET
241 #include <net/dst.h>
242 
243 /* The function in 2.2 was invalid, producing wrong result for
244  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
245 static inline
246 int ip_decrease_ttl(struct iphdr *iph)
247 {
248 	u32 check = (__force u32)iph->check;
249 	check += (__force u32)htons(0x0100);
250 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
251 	return --iph->ttl;
252 }
253 
254 static inline
255 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
256 {
257 	return  inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
258 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
259 		 !(dst_metric_locked(dst, RTAX_MTU)));
260 }
261 
262 void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more);
263 
264 static inline void ip_select_ident(struct sk_buff *skb, struct dst_entry *dst, struct sock *sk)
265 {
266 	struct iphdr *iph = ip_hdr(skb);
267 
268 	if ((iph->frag_off & htons(IP_DF)) && !skb->local_df) {
269 		/* This is only to work around buggy Windows95/2000
270 		 * VJ compression implementations.  If the ID field
271 		 * does not change, they drop every other packet in
272 		 * a TCP stream using header compression.
273 		 */
274 		iph->id = (sk && inet_sk(sk)->inet_daddr) ?
275 					htons(inet_sk(sk)->inet_id++) : 0;
276 	} else
277 		__ip_select_ident(iph, dst, 0);
278 }
279 
280 static inline void ip_select_ident_more(struct sk_buff *skb, struct dst_entry *dst, struct sock *sk, int more)
281 {
282 	struct iphdr *iph = ip_hdr(skb);
283 
284 	if ((iph->frag_off & htons(IP_DF)) && !skb->local_df) {
285 		if (sk && inet_sk(sk)->inet_daddr) {
286 			iph->id = htons(inet_sk(sk)->inet_id);
287 			inet_sk(sk)->inet_id += 1 + more;
288 		} else
289 			iph->id = 0;
290 	} else
291 		__ip_select_ident(iph, dst, more);
292 }
293 
294 /*
295  *	Map a multicast IP onto multicast MAC for type ethernet.
296  */
297 
298 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
299 {
300 	__u32 addr=ntohl(naddr);
301 	buf[0]=0x01;
302 	buf[1]=0x00;
303 	buf[2]=0x5e;
304 	buf[5]=addr&0xFF;
305 	addr>>=8;
306 	buf[4]=addr&0xFF;
307 	addr>>=8;
308 	buf[3]=addr&0x7F;
309 }
310 
311 /*
312  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
313  *	Leave P_Key as 0 to be filled in by driver.
314  */
315 
316 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
317 {
318 	__u32 addr;
319 	unsigned char scope = broadcast[5] & 0xF;
320 
321 	buf[0]  = 0;		/* Reserved */
322 	buf[1]  = 0xff;		/* Multicast QPN */
323 	buf[2]  = 0xff;
324 	buf[3]  = 0xff;
325 	addr    = ntohl(naddr);
326 	buf[4]  = 0xff;
327 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
328 	buf[6]  = 0x40;		/* IPv4 signature */
329 	buf[7]  = 0x1b;
330 	buf[8]  = broadcast[8];		/* P_Key */
331 	buf[9]  = broadcast[9];
332 	buf[10] = 0;
333 	buf[11] = 0;
334 	buf[12] = 0;
335 	buf[13] = 0;
336 	buf[14] = 0;
337 	buf[15] = 0;
338 	buf[19] = addr & 0xff;
339 	addr  >>= 8;
340 	buf[18] = addr & 0xff;
341 	addr  >>= 8;
342 	buf[17] = addr & 0xff;
343 	addr  >>= 8;
344 	buf[16] = addr & 0x0f;
345 }
346 
347 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
348 {
349 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
350 		memcpy(buf, broadcast, 4);
351 	else
352 		memcpy(buf, &naddr, sizeof(naddr));
353 }
354 
355 #if IS_ENABLED(CONFIG_IPV6)
356 #include <linux/ipv6.h>
357 #endif
358 
359 static __inline__ void inet_reset_saddr(struct sock *sk)
360 {
361 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
362 #if IS_ENABLED(CONFIG_IPV6)
363 	if (sk->sk_family == PF_INET6) {
364 		struct ipv6_pinfo *np = inet6_sk(sk);
365 
366 		memset(&np->saddr, 0, sizeof(np->saddr));
367 		memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
368 	}
369 #endif
370 }
371 
372 #endif
373 
374 static inline int sk_mc_loop(struct sock *sk)
375 {
376 	if (!sk)
377 		return 1;
378 	switch (sk->sk_family) {
379 	case AF_INET:
380 		return inet_sk(sk)->mc_loop;
381 #if IS_ENABLED(CONFIG_IPV6)
382 	case AF_INET6:
383 		return inet6_sk(sk)->mc_loop;
384 #endif
385 	}
386 	WARN_ON(1);
387 	return 1;
388 }
389 
390 bool ip_call_ra_chain(struct sk_buff *skb);
391 
392 /*
393  *	Functions provided by ip_fragment.c
394  */
395 
396 enum ip_defrag_users {
397 	IP_DEFRAG_LOCAL_DELIVER,
398 	IP_DEFRAG_CALL_RA_CHAIN,
399 	IP_DEFRAG_CONNTRACK_IN,
400 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
401 	IP_DEFRAG_CONNTRACK_OUT,
402 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
403 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
404 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
405 	IP_DEFRAG_VS_IN,
406 	IP_DEFRAG_VS_OUT,
407 	IP_DEFRAG_VS_FWD,
408 	IP_DEFRAG_AF_PACKET,
409 	IP_DEFRAG_MACVLAN,
410 };
411 
412 int ip_defrag(struct sk_buff *skb, u32 user);
413 #ifdef CONFIG_INET
414 struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user);
415 #else
416 static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user)
417 {
418 	return skb;
419 }
420 #endif
421 int ip_frag_mem(struct net *net);
422 int ip_frag_nqueues(struct net *net);
423 
424 /*
425  *	Functions provided by ip_forward.c
426  */
427 
428 int ip_forward(struct sk_buff *skb);
429 
430 /*
431  *	Functions provided by ip_options.c
432  */
433 
434 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
435 		      __be32 daddr, struct rtable *rt, int is_frag);
436 int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
437 void ip_options_fragment(struct sk_buff *skb);
438 int ip_options_compile(struct net *net, struct ip_options *opt,
439 		       struct sk_buff *skb);
440 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
441 		   unsigned char *data, int optlen);
442 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
443 			     unsigned char __user *data, int optlen);
444 void ip_options_undo(struct ip_options *opt);
445 void ip_forward_options(struct sk_buff *skb);
446 int ip_options_rcv_srr(struct sk_buff *skb);
447 
448 /*
449  *	Functions provided by ip_sockglue.c
450  */
451 
452 void ipv4_pktinfo_prepare(struct sk_buff *skb);
453 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
454 int ip_cmsg_send(struct net *net, struct msghdr *msg, struct ipcm_cookie *ipc);
455 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
456 		  unsigned int optlen);
457 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
458 		  int __user *optlen);
459 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
460 			 char __user *optval, unsigned int optlen);
461 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
462 			 char __user *optval, int __user *optlen);
463 int ip_ra_control(struct sock *sk, unsigned char on,
464 		  void (*destructor)(struct sock *));
465 
466 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
467 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
468 		   u32 info, u8 *payload);
469 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
470 		    u32 info);
471 
472 #ifdef CONFIG_PROC_FS
473 int ip_misc_proc_init(void);
474 #endif
475 
476 #endif	/* _IP_H */
477